Cite this: DOI: 10.1039/c3ra42205d

Compositional characterisation of metallurgical gradesilicon and porous silicon nanosponge particles

Received 3rd May 2013,Accepted 1st August 2013

DOI: 10.1039/c3ra42205d

www.rsc.org/advances

Edward G. Chadwick,ab N. V. V. Mogili,ab Colm O’Dwyer,3ac Jimmy D. Moore,d JohnS. Fletcher,{d Fathima Laffir,a Gordon Armstronga and David A. Tanner*ab

Porous silicon is generally achieved through electro-chemical etching or chemical etching of bulk silicon in

hydrofluoric acid based solutions. The work presented here explores the effect of a chemical etching

process on a metallurgical grade silicon powder. It is found that the metallurgical grade silicon particles

contain surface bound impurities that induce a porous structure formation upon reaction with the

chemical etchant applied. The correlation between the resultant porous structure formed due to the

material composition is examined in detail. The elemental composition is determined using a combination

of X-ray Photoelectron Spectroscopy and Time of Flight Secondary Ion Mass Spectroscopy. The porous

structure is analysed using Transmission Electron Microscopy and Scanning Electron Microscopy. Three

samples of the silicon particles analysed for this study include an un-etched bulk silicon powder sample

and two samples of chemically etched powder. Pore formation within the particles is found to be

dependent on the presence, dispersion, and local concentration of surface bound impurities within the

starting powder.

1. Introduction

Porous silicon (PS) was originally discovered in 1956 by ArthurUhlir and Ingeberg Uhlir at the Bell Laboratories in the USwhen electropolishing silicon (Si) wafers using hydrofluoricacid (HF).1 What formed was a coloured porous film on the Siwafer with pore formation occurring in the [100] direction.2

While this find was noted there was little interest in PS until1990 when Canham and Lehman et al. made the discovery ofits room temperature photoluminescence as a result ofquantum confinement effects.3–7 There has since been anabundance of research into porous silicon and relatedmaterials6,8–12 to develop possible applications in areas suchas optoelectronics, microelectronics, chemical and biologicalsensors, drug delivery devices and more recently energeticmaterial devices.13,14 PS therefore has the potential fornumerous applications in a variety of fields but extensive

characterisation of the material is deemed essential beforepursuing possible Si material applications development.

Metallurgical grade Si (MGSi) contains a host of impuritiessuch as Fe, Al, Ca, C and O.15,16 These impurities are located atgrain boundaries of the material when it is being manufac-tured.16 Vesta Sciences have developed a process for producinglow cost PS from MGSi powders through a patented chemicaletching process.15,17 In this work, a detailed compositionalinvestigation is carried out on this MGSi and the metallurgicalgrade porous silicon (MGPS) produced post-chemical etchingin line with a structural characterisation of the particleporosity. The relationship between the composition of thematerial and the resulting porous structure formation isexamined in detail in order to understand the porosityformation mechanisms of this material. It is shown that theporous morphology is that of a quantum sponge particle withSi crystallites arranged in a random porous matrix. Detailedelemental investigation links the specific composition of thePS produced from the process to the resulting mechanism ofpore nucleation and formation. These results provide aninsight into the development of this nanosponge materialwhen processed under such conditions. This insight can beused to help further tailor the material for various applicationsfrom biomedical engineering to military applications.Microscopical techniques including Transmission ElectronMicroscopy (TEM) and Convergent Beam Electron Diffraction(CBED) were applied to the material. Characteristics includingpore size, volume and pore orientation are studied as an effectof the processing parameters and etching mechanism, where

aMaterials and Surface Science Institute, University of Limerick, Limerick, Ireland.

E-mail: david.tanner@ul.ie; Fax: 00-353-(0)61-202913; Tel: 00-353-(0)61-234130bDepartment of Design and Manufacturing Technology, University of Limerick,

Limerick, IrelandcDepartment of Physics and Energy, University of Limerick, Limerick, IrelanddSchool of Chemical Engineering and Analytical Science, Manchester

Interdisciplinary Biocentre, University of Manchester, 131 Princess St., Manchester,

M1 7DN, UK

3 Present address: Department of Chemistry, University College Cork, Cork,Ireland, and Tyndall National Institute, Lee Maltings, Cork, Ireland(c.odwyer@ucc.ie).{ Present address: Department of Chemistry and Molecular Biology, Universityof Gothenburg, 41296 Gothenburg, Sweden.

RSC Advances

PAPER

This journal is � The Royal Society of Chemistry 2013 RSC Adv.

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

.

View Article OnlineView Journal

the elemental composition was determined using X-rayPhotoelectron Spectroscopy (XPS) and Time of FlightSecondary Ion Mass Spectroscopy (TOF-SIMS). Raman spectro-scopy was conducted to examine variations in crystallitestructure and strain in PS arising from changes to theconditions used to prepare the PS.

2. Experimental

2.1 Porous silicon preparation

The MGSi powder is supplied from Vesta Ceramics (Sweden)which is sold under the trademark of SicomillTM. The powderinitially has a particle size of about 1–3 mm before being jetmilled. The jet milled particles are then classified into severalgrades with different mean particle sizes ranging from 4 mm to11 mm. For the study here, E grade particles with a meanparticle size of 4 mm are used. PS particles are prepared bychemically etching the E grade MGSi particles using apatented etching process carried out by Vesta Sciences.15,17

The particles are chemically etched in a nitric-acid (NO3)–(HF)based mixture as described by Farrell et al.15 which inducesthe porosity within the MGSi particles. When the etchingprocess is complete (i.e. photoluminescence is observed due toquantum confinement of Si feature sizes below the Bohrradius of the material and also from defect induced emissionfor features .4–5 nm in size), the etched powder is removedfrom the acid bath and dried in perfluoroalkoxy (PFA) trays at80 uC for 24 h. By varying the total nitric acid concentration inthe etchant mixture the surface area of these powders can alsobe tailored.15,17 The three samples of MGSi powders analysedin this paper are defined as follows (‘2E’ refers to a specificgrade of material):

1) 2E-Bulk – the starting MGSi powder used to make the PSparticles.

2) 2E-100 – PS powder manufactured from the 2E-Bulkpowder after chemically etching with a nitric acid concentra-tion corresponding to that needed to obtain the highestsurface area and photoluminescence (100% is defined as thenitric acid concentration that results in powders that exhibitthe higher surface area, porosity and photoluminescence withan ultraviolet light15).

3) 2E-60 – as for (2) but only chemically etched with 60% ofthe nitric acid used for 2E-100.

2.2 High resolution electron microscopy

TEM specimens were prepared by loading the PS particles ontoFormvar-backed carbon-coated copper grids (Agar Scientific).TEM analysis was acquired using a JEOL JEM 2100F FieldEmission Electron Microscope operating at 200 kV.

TEM images were recorded with a Gatan Ultrascan 1000digital camera for standard TEM imaging modes. The TEM isalso equipped with a secondary electron imaging (SEI) detectorwhich allows Scanning Electron Microscopy (SEM) type imagesto be recorded on the TEM. For CBED analysis, crystals weretilted to the [340] zone axis and a series of CBED patternsrecorded in Scanning Transmission Electron Microscopy

(STEM) mode with a spot size of 1.5 nm using STEM-baseddiffraction imaging.18

2.3 Time-of-flight secondary ion mass spectrometry

Time of Flight Secondary Ion Mass Spectroscopy (ToF-SIMS)analysis of the 2E-bulk and etched 2E-100 Si sample wasperformed using the J105 3D Chemical Imager (Ionoptika LtdUK). A layer of each sample was deposited onto a siliconsubstrate from a dense suspension in methanol. Images wereacquired over a 500 6 500 mm2 field of view, with a primaryion dose of 1.7 6 1013 ions/cm2 using a 40 keV C60

+ primaryion beam. Following acquisition principal components analy-sis (PCA) was performed on the two data sets. PCA is amultivariate technique commonly used in SIMS and is a widelyused dimensionality reduction technique for data analysis andinterpretation. PCA converts the data into a series of principalcomponents (PCs). Each PC has an associated loading which isthe contribution of each mass channel to that PC. Each pixel/spectrum then has a corresponding score on that loading.Principal components are ordered highest to lowest (highestbeing PC1) in terms of the degree of variance within the datacaptured by that PC.

2.4 Raman scattering spectroscopy

Raman scattering spectra were acquired from powder disper-sions of both as-received and etched MGSi particles using aHoriba Jobin Yvon Labram 1A Raman spectrometer. Spectrawere collected using a 532 nm 100 mW laser source, 1800 line/mm grating and 106 objective, unless noted otherwise. Thespectrometer was calibrated against the 521 cm21 TransverseOptical (TO) phonon frequency of a Si (100) wafer; theresolution of the spectrometer is , 2 cm21. Each spectrumpresented is the average of three accumulations.

2.5 X-ray photoelectron spectroscopy

XPS was performed to analyse the surface chemistry of the Siparticles using a Kratos AXIS 165 spectrometer with monochro-matic Al Ka radiation of energy 1486.6 eV. High resolutionspectra were taken at fixed pass energy of 20 eV. Binding energieswere determined using C 1s peak at 284.8 eV as charge reference.For construction and fitting of synthetic peaks of high resolutionspectra, a mixed Gaussian–Lorenzian function with a Shirley typebackground subtraction were used.

2.6 X-ray fluorescence and inductively coupled plasmaspectroscopy

The samples were supplied with X-Ray Fluorescence (XRF)Spectroscopy and Inductively Coupled Plasma (ICP)Spectroscopy completed by Vesta Sciences as shown in section4.1.

3. Experimental results and discussion –morphology

3.1 SEM and TEM of silicon particle surface structure

The particle morphology was studied from a series of SEM andTEM images. Fig. 1a and b show high resolution SEI images of

RSC Adv. This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

sample 2E-100. These images were acquired in SEI mode,which provided higher resolution imaging of the surfacemorphology compared to FESEM and thus shows the poroussurface area in much more detail. On examination of theselected PS particle it is evident that the surface of the particleis completely porous in nature. The Si particle size ranges from4–5 mm with pore sizes ranging from 8–15 nm in diameter.19

Similar porosity was observed for sample 2E-60 but with asmaller pore volume per surface area.19 The 2E-bulk Si sampleconsists of solid Si particles.19

3.2 TEM-CBED of silicon particles

TEM-CBED analysis was performed on the surface of the2E-Bulk sample to examine the possible strain change withinthe material when compared to sample 2E-100. SecondaryElectron Imaging (SEI) in combination with high spatiallyConvergent Beam Electron Diffraction (CBED) was performed.Due to the converged beam of illumination, High Order LaueZone (HOLZ) lines are formed inside the CBED patterns(Fig. 2) and are very sensitive to any lattice changes. Based onthe appearance of HOLZ lines, the qualitative nature of thepore structure was determined.

Along the spectrum image line as indicated in Fig. 3a for anunetched Si particle (2E-Bulk), successive [340] HOLZ lines arecollected to identify any local lattice distortions. ,004. two

beam CBED patterns are also recorded along the spectrumimage line and based on the distance between Kossel–Mollenstedt (KM) fringes inside CBED discs, particle thicknesshas been calculated and shown to vary from 358 nm to 393nm. Fig. 3b and c show two experimental [340] HOLZ linepatterns obtained from the spectrum image series. Fig. 3d ande highlight the region where the HOLZ lines intersect showingno shift in the HOLZ lines intersection point indicating thatthe lattice spacing remains constant within this region.

In the fully etched porous Si particles (2E-100), a series of[340] HOLZ line patterns were taken along the spectrum imageline as seen in Fig. 4a. Fig. 4b and c show two different HOLZlines patterns that were part of a spectrum imaging seriesclearly displaying a shift in the HOLZ line intersectionpositions; the shift is shown at higher magnification inFig. 4d and e. As the HOLZ lines are related to the HOLZreflections with large reciprocal lattice vectors of the crystallattice, their positions are highly sensitive to lattice deforma-tions.20 To further understand the HOLZ line shift diffractioncontract imaging is performed as discussed in section 3.3

3.3 Diffraction contrast imaging of porous silcion nanospongeparticles

Dark field HRTEM imaging is used in a two-beam condition toobtain high contrast images of the crystalline phases. Fig. 5,represents ,220. dark field images obtained from (a) sample2E-bulk and (b) 2E-100. The required diffraction contrast wasestablished by tilting the particles from the [011] zone axisorientation towards the (220) reflection. In Fig. 5a, due to thenon-uniform thickness of the Si particle in the electron beamdirection, thickness fringes are observed. In Fig. 5b, brightspots are observed on the particle surfaces indicating that

Fig. 1 (a) SEI micrograph of sample 2E-100 oriented in [340] zone axis (b)Corresponding higher magnified image clearly showing porous nature of theparticle.

Fig. 2 a) Convergent Beam Imaging (CBIM) micrograph of [340] orientedunetched Si particle exhibiting both real and reciprocal space information b)Corresponding Secondary Electron Image (SEI) of unetched Si particle illustrat-ing the surface morphology of the crystal.

Fig. 3 a) SEI image of a 2E-bulk particle indicating the spectrum image linewhere sequential CBED patterns are collected b), c) Two [340] CBED patternstaken from spectrum image and HOLZ lines intersection points which are verysensitive for lattice changes are highlighted d), e) Enlarged region of [340] HOLZlines showing that there is no shift/change in intersection point.

This journal is � The Royal Society of Chemistry 2013 RSC Adv.

RSC Advances Paper

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

crystalline Si nanoparticles appear on the surface after etching.These Si clusters have been etched out of the bulk Si particlesand are scattered among the porous regions of the PS samples.Hence, from the dark field imaging, it is clear that Si nano-crystallites exist on the PS particle surface, so the main causefor the line-shift observed in the PS particles in section 3.2 isdue to presence of Si nano-crystallites on the particle surfaces.

The analysis here corresponds to work by Peng et al.describing the mechanism for pore formation showing metal

particles at the bottom of the pores of a Si sample materialafter etching and pore formation.21 In the case of MGPS it isproposed that the metallic impurities which are on the surfacein the 2E-bulk sample could also be driven into the Si materialduring pore formation leaving only the removed Si dispersedamong the surface pores, confirming the relative abundancein variations between elements detected by XRF, ICP and TOF-SIMS post-etching described in later sections. Si nano-particleshave also been observed previously where particles weredispersed among a surface bound oxide layer post-electro-chemical etching.22

4. Experimental results and discussion –elemental analysis

4.1 X-ray fluorescence and inductively coupled plasmaspectroscopy of silicon particles

XRF and ICP analysis were carried out on samples 2E-Bulk and2E-100 to explore the concentrations of elements presentaround the surface and within the Si particles. Table 1 showsthe quantitative concentrations of elements detected. It wasfound that the Fe (0.5%) and Al (0.2) concentrations aregreatest, followed by Ca (0.08%) and C (0.06%) in sample2E-Bulk. These values indicate just how small these levels ofimpurities are dispersed throughout the Bulk Si particles.These values correspond to the patent standard and are also inline with what is observed for standard MGSi powder.15,16 Thelevels of Fe, Al, Ca, C and O observed in sample 2E-100 usingXRF analysis are less than those observed in sample 2E-Bulk.This corresponds to the mechanism that the MGSi powdercontains impurities which, upon reaction with the HF basedelectrolyte, induce porosity at the Si particle surface viaelectroless etching processes. The impurities are then attackedin a cathodic reaction which leaves a porous structurethroughout the outermost region of the Si particle. Theseimpurities would then be significantly reduced in the materialto smaller amounts dispersed throughout the remaining non-porous region of the PS sample. Previous reports have shownthat the application of certain elements onto a Si surface willinduce porous layer formation once electro-chemically orchemically etched in a HF based solution. Metal-assistedetching using thin films of Pt and Ag on Si has shown PSformation23–26 and incorporating Fe into HF solutions has alsobeen shown to affect the etching process to some degree.27 Alhas already been shown to induce a porous layer structureformation in Si in a study which chemically etched a thin Alfilm on a Si substrate.28 The Al film (up to 200 nm inthickness) reacted with the HF solution and in this case wasthe key mechanism for hole initiation and the porousstructure formation.28 These results correlate directly to theporous structure formation mechanism proposed in this paperand is further highlighted by the mass loss (% reductions) ofthe impurities pre and post chemical etching. Fe, which is themost abundant element, is reduced by 84% while Al and Caare reduced by 81% and 83% respectively.

Fig. 4 a) High resolution SEI micrograph of sample 2E-100 PS sample showingspectrum imaging location of one particular pore b), c) Two [340] CBED patternscollected from spectrum image showing the shift of HOLZ line position d), e)Enlarged region of [340] HOLZ lines clearly showing shift in the position ofintersection point. (Reader is advised to go through the supplementary videofile in order to visualise how the HOLZ lines move in the spectrum imagingpatterns).

Fig. 5 (a) Typical ,220. dark field images of sample 2E-bulk Si particle and (b)sample 2E-100 PS particle.

RSC Adv. This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

4.2 X-ray photoelectron spectroscopy of silicon particles

XPS was utilised to establish the surface layer composition ofthe PS particles. The three samples analysed included the2E-bulk Si sample, sample 2E-100 and sample 2E-60. Allsamples showed the presence of Si, O and C and in the case ofchemically etched 2E-100 and 2E-60 samples notable amountsof F were also identified. Other impurities were not observedperhaps due to their very low concentrations (less than thedetection limit of XPS y0.1 atomic%). Fig. 6 shows the XPS

survey spectra of the 2E-bulk Si sample. The prominent peaksare Si (Si 2p, 2s), C (C 1s) and O (O 1s, KLL X-Ray inducedAuger transition). These peaks are similar to previous XPSanalysis of PS.29 Feng et al. has shown that 200 mm thick PSmembranes showed similar peak positions to those observedhere for the PS and Si particles.29 Quantification of elementsfrom XPS analysis is given in Table 2 with relative percentageconcentrations of F at 2%, C at 28.6%, Si at 60.2%, and O at9.2% for 2E-100. Similar results were obtained for 2E-60. Therespective peaks for 2E-bulk Si indicated percentage concen-

Table 1 XRF and ICP concentrations of surface elements for 2E-Bulk and 2E-100a

Elements XRF (2E-Bulk) % ICP (2E-Bulk) % XRF (2E-100) % (% XRF Reductions)

Fe 0.3–0.5 0.31 0.08 73–84Al 0.08–0.2 0.18 0.038 53–81Ca 0.03–0.08 0.031 0.014 53–83C ,0.06 0.0327¡0.0010 0 100O 0 0.3722¡0.0038 0 0

a All results were obtained in collaboration with Vesta Research.

Fig. 6 a) XPS analysis of sample 2E-bulk; b) and c) high resolution Si 2p spectra of 2E-bulk and 2E-60 respectively.

This journal is � The Royal Society of Chemistry 2013 RSC Adv.

RSC Advances Paper

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

trations C at 8.7%, Si at 61.1%, and O at 30.2%. The presenceof F in the PS samples can be attributed to the etching process.

A high resolution Si 2p spectrum from the 2E-bulk Si sampleis shown in Fig. 6b and can be fitted with three sets ofdoublets. The majority of silicon is present as elemental Siwith Si 2p3/2 at a binding energy of 98.8 eV. The peaks athigher binding energy of 101.1 eV and 102.8 eV can beattributed to silicon oxide SiOx (x , 2) and SiO2 respec-tively.30,31 The presence of the related O 1s at 532.3 eVconfirms that the more abundant of the oxides, SiO2, is largelystoichiometric. However, Si 2p spectra of sample 2E-60(Fig. 6c) are dominated by low binding energy peakscorresponding to elemental silicon with a relatively minorcontribution from higher binding energy components at 103.6eV which may be attributed to SiO2 or a residual SiFx complexfrom the etching process, which proceeds according to a seriesof galvanic displacement reactions according to

M+ + e2 A M0 (where M is the metal impurity)

Si + 2H2O A SiO2 + 4H+ + 4e2

SiO2 + 2HF22 + 2HF A SiF6

22 + 2H2O

The higher content of oxygen on the surface of sample2E-bulk (30.2 atomic%) can be attributed to it being the bulkstarting material and likely to have undergone surfaceoxidation. The O content is significantly reduced in the PSsamples indicating it was removed through the appliedchemical etching process and HF reaction at the Si surface.Increased carbon content on the surface of the etched PS isdue to the environment in which XPS analysis is conducted.

4.3 Time of flight secondary ion mass spectroscopy of siliconparticles

Further analysis of the Si particle composition was carried outusing TOF-SIMS analysis. Sample 2E-bulk showed the standardelements associated with the MGSi powder. The total ionimages of both the bulk and etched samples can be seen inFig. 7. Images were acquired over a 500 6 500 mm2 field ofview, with a primary ion dose of 1.7 6 1013 ions cm22 using a40 keV C60

+ primary ion beam.The scores and loadings for the Principal Component (PC) 2

of the 2E-bulk un-etched particles can be seen in the spatialscore maps in Fig. 8a. The scores image (Fig. 8a) highlightswhere red indicates a negative score and green indicates apositive score. In comparison to the total ion image (Fig. 7a) it

is apparent that the highlighted regions with this chemistryi.e. green regions are cluster/groups of the Si particles. Fromthe corresponding loadings (Fig. 8b) it is evident that the Al+,Ca+ and Fe+ are correlated to these Si cluster groups,confirming the elemental composition analysis from XRFand ICP analysis in Table 1. Some salts (Na+, K+) are alsocorrelated in this PC.

The etched samples (2E-100) can be considered in a similarmanner. The scoring map indicates a reduced abundance of theAl+, Ca+ and Fe+ species, captured in the PC3 loading in Fig. 8d.These analyses have demonstrated the presence of these metalson the Si particles before and after etching and are thus notreleased into the etching solution after chemical reductionduring the etching process. Remnant Na+, K+ were easilydissolved and removed from the particles in the aqueous HF-based etching solution. However, changes in the valence orquantitative relative abundances cannot be deciphered from thisanalysis, particularly since the presence of reduced metal speciesafter galvanic displacement or electroless reactions are lodgedwithin the nanoporous structure of the silicon sponge particles.

4.4 Raman scattering spectroscopy of porous silicon quantumsponge particles

Raman spectroscopy is a powerful technique that can beemployed in structural and dynamic studies of PS.29,32 A typicalRaman spectrum for crystalline Si usually consists of asymmetric first order transverse optical phonon centred aty521 cm21.33,34 As the grain size in micro-crystalline Sidecreases, the Raman shift and peak width increases and theline-shape becomes asymmetric (which also occurs with dopingchanges, which is invariant in this case) with an extended tail atlower frequencies for quantum confinement or small sizeeffects. In amorphous Si the Raman peak is broad and usuallyweak at 480 cm21.33,34 Here, micro-Raman scattering spectro-scopy was conducted on samples 2E-100, 2E-60 and 2E-bulk Si(Fig. 9a) to examine the development and influence of porosityon the overall structure, specifically the presence and associatedinfluence of nano-sized crystallites of silicon.

The main Si peak in the 2E-bulk Si occurs at 517 cm21,compared to 521 cm21 measured for bulk Si(100) and the TOphonon mode is sharp and symmetric compared to highlydoped Si(100), which displays a blue asymmetry indicative ofhighly p-type Si. The TO modes for both 2E-60 and 2E-100

Table 2 XPS % concentrations of surface elements for 2E-100, 2E-60 and2E-bulk

% Concentrations Sample 2E-100 Sample 2E-60 Sample 2E-bulk

Fluorine 2.0 1.3 0Carbon 28.6 24.5 8.7Silicon 60.2 65.6 61.1Oxygen 9.2 8.6 30.2

Fig. 7 Total ion images (500 6 500 mm2) of (a) 2E-bulk sample (b) 2E-100particles.

RSC Adv. This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

occur at 516 cm21 and 510 cm21 (Fig. 9b), has an asymmetricline shape with asymmetry appearing on the lower (red) energyside indicative of phonon confinement due to reduced sizes,as the doping density remains invariant for the same particlesand no excess heating has been applied externally or throughJoule heating by the incident Raman probe beam. Generallyfor PS it is considered that the broadening and a downshift ofthe TO mode towards lower energies indicates the presence ofnanoscale features of the crystalline structures and is distinctfor high porosity layers of PS.35 A slight lower energy downshiftof the peaks was observed in 2E-100 compared to 2E-bulk Si.These quantum confined features come from the smallestpore wall structures but particularly from the high density ofsmall ,5 nm crystallites seen by STEM, as outlined in section3.3. These core Si values do not correlate directly with Si or PSsubstrates with typical Raman peaks around 521 cm21 but aremore closely linked to the PS membranes studied by Fenget al.29 In that work, Raman spectra of PS membranes removedfrom anodically oxidised p-type ,100. Si substrates, showeddominant Raman shifts between 502 and 512 cm21.29 Thebands located near 300 cm21 correspond to an acousticphonon (2TA). These values correlate directly to what isobserved for sample 2E-100 in Fig. 9. The features between900–1000 cm21 indicate second order optical modes,29 as seenin both samples 2E-100 and 2E-60.

It has also been shown that surface strain or stress in PS canalso contribute to a downshift in the Raman peak as isobserved from 2E-bulk Si to 2E-100 in Fig. 9b. The positions ofthe characteristic Raman peaks of a material correspond to thevibrational energy of the phonon modes. However, if thecrystal lattice deforms, the position of the mode may shift tothe left or to the right, due to an energy variation in the

vibration of the lattice. Lang et al. reports that the latticeconstant of PS is somewhat strained and a stress valuecorresponding to such a lattice strain is known to shift theRaman peak to lower energies36 known as phonon softening.The incorporation of a surface oxide may relax the surfacestrain to some degree in PS.36 The authors discuss a linearrelationship between uniaxial or biaxial stress and the shift ofthe TO mode of the sample. The reduction in total free energywithin the particle, is such that the porous particle is renderedmore brittle compared to the bulk particle can be calculatedfrom the empirical formula37e = sst?Dv, where v representsthe peak positions in cm21 for strained and bulk Si, and Dv

represents (vbulk 2 vpore), sst = 0.123 is the inverse strain-phonon coefficient in cm, and e is the % strain.

The quantum confinement caused by crystallites of siliconand also the smallest of porous wall features by monitoringthe asymmetry and frequency shift or the TO phonon mode isshown in Fig. 9b. For both 2E-bulk and 2E-100 samples, theRaman shift from respective TO modes correlates to a % freeenergy reduction of 47% for 2E-bulk particles compared to a Si(100) crystal, and a further 18% reduction (65% compared toSi(100)) upon partial porosification for 2E-60 samples, but avery large 136% reduction in free energy for fully etched 2E-100particles compared to Si(100). In this process, every stressrelieving interface will most likely contain metallic ionsamenable to electroless etching and thus all galvanicdisplacement reactions on the surface will lead to a significantreduction in internal particle free energy.

The crystallite size L of the porous Si quantum spongeparticles can be estimated from the Raman shift according to:

Dv = 52.3 (0.543/L)a

Fig. 8 (a, b) Scores image and the corresponding TOF-SIMS loading spectrum from PC2 of the 2E-bulk Si particles and (c, d) from PC3 of the etched 2E-100 Si particles.

This journal is � The Royal Society of Chemistry 2013 RSC Adv.

RSC Advances Paper

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

where Dv is the Raman shift between that from a silicon wafer(521 cm21) and the TO phonon frequency from each of the TOphonon peaks from the Si sponge particles. The power lawdependence determined empirically is a = 1.59. The crystallitesizes for 2E-60 and 2E-100 (with respect to a Si wafer) are estimatedto be 1.93 and 1.16 nm respectively. These values correspond tothe electron microscopy observations in Section 3.3.

5. Discussion – porosity formation

5.1 Porosity formation in metallurgical grade silicon

Understanding the porosity formation mechanisms in Si,when processed under certain conditions is essential forsuccessful tailoring of the material for final applications.Hence there has been an abundance of work completed ondifferent Si materials using different etching methods.38–41 PSis generally achieved when doped Si wafers are electrochemi-cally or chemically etched using HF based solutions.42,43 Pore

formation and propagation is usually dependant on thecrystallographic orientation of the Si wafers used in anelectrochemical or chemical etching process. Other importantproperties of the PS material such as resultant pore size, porevolume and pore depth are very much dependant on theanodisation conditions applied i.e. HF concentration, oxida-tion, carrier type (n-type or p-type), current density applied, theanodisation method and duration etc.8–10,44,45 These para-meters can be tuned to some degree resulting in tailoring ofthe Si porosity to desired conditions. Other etching methodsderived from standard electrochemical and chemical etchingprocesses include metal-assisted etching and in this method,PS is produced by electroless etching and galvanic displace-ment reactions between certain metallic ions at the surface ina fluoride-containing electrolyte.23–26 The application of a thinmetallic film or metal particles (Au, Pt, Al, Ag, palladium (Pd))onto the surface of the Si substrate results in a cathodic–anodic reaction (metal film or particle being the cathode) andthe Si material (when oxidised becoming the anode) whichthen results in a boring or sinking of the metal particles intothe Si surface creating pore formation. Various metallic–Siinterface reactions can be explained but essentially in the caseof Ag particles for example, the Si beneath the Ag is oxidised byhole injection when the oxidising agent (hydrogen peroxide(H2O2), sodium persulfate (Na2S2O8) or HNO3 are typicallyused in the etching solution) is reduced on the Ag. Thereforethe Si is oxidised which reacts with the HF to form a water-soluble complex and thus the holes are then simply thesinking or boring tracks of the Ag particles.24 Etching pure Siwafers in HF without prior hole injection simply results in apassivated surface layer.46 Understanding these porosityformation mechanisms is key in outlining the mechanismsresponsible for porosity formation in MGSi containingmetallic impurities, processed under similar conditions.

XRF and ICP analysis confirmed the presence of Fe, Al, Ca, Cand O dispersed throughout the MGSi particles. These metalions which are considered mostly surface bound within the Siparticles react with the HF based chemical etchant applied,which in the case of this study is HF, HNO3 and water (H2O).Nitric Oxide (NO) is produced on the surface of the Si whichserves as the hole injector.44 The rate of dissolution dependson the number of injected holes per atom, which in thepresent case is greatest for Al3+ compared to Fe2+ or Ca+. Beckand Gerischer proved that the reaction rate is proportional tothe surface concentration of holes.47 Due to the differentreactivities of surface electrons and holes, the doping type ofthe crystal leads to very different reactivites for n-type andp-type doping; in the case of intrinsically doped MGSi particleshere, the role of the hole injector ions is paramount.

The presence of a hole in the valence band reduces thestrength of bonds in the vicinity and makes the substrateatoms more susceptible to attack by nucleophiles. Similarly,the presence of a conduction band electron weakens bonds inits vicinity and makes those substrate atoms susceptible toelectrophilic attack. Additionally, the likelihood for thesegalvanic displacement reactions also depends on having asmall electronegativity difference between the metal ion andthe silicon and coupled with the relative abundance of therespective ions, leads to variable etching rates. All three will

Fig. 9 (a) Micro-Raman spectrum of PS samples 2E-100, 2E-60 and 2E-bulk Si (b)TO phonon modes for PS samples 2E-100, 2E-60 and 2E-bulk Si.

RSC Adv. This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

participate and their spatial density on the particle (highest onthe surfaces which were originally the grain boundaries in thepre-milled MGSi) results in a complicated process where theresult is a Si ‘sponge’. It is noted that the Si ‘sponge’ stillcontains some levels of impurities which are greatly reducedas determined using XRF and ICP. The TOF-SIMS spectracorrelate well with the XRF and ICP analyses when the galvanicdisplacement reactions are considered. Each of the metal ionsis reduced by extracting a number of electrons from the siliconequivalent to the valence of the ion; the rate of electrolessetching is greatest for Al and thus the relative abundance isminimum post-etching as the metallic Al0 is containedfurthest into the sponge structure (assuming some of eachelement is lost through equivalent mass transport into theaqueous etching solution). This etching process also has aresultant effect in that Si remnants are removed during theetching process and are then dispersed among the surfacepores as nano-crystallites. This is observed through TEM-CBED and Diffraction contrast imaging of the PS particles andconfirms the relative abundance in variations betweenelements detected by XRF, ICP and TOF-SIMS post-etching.Raman Spectroscopy confirmed these Si nano-crystallitesresult in a lower energy downshift of the peaks as observedin 2E-100 compared to 2E-Bulk Si and the calculated crystallitesize conforms to what is observed through diffraction contrastimaging. The Raman data points to a reduction in free energywithin the samples after etching, which may indicate strainrelaxation within individual particles as a result of thereduction of impurity levels.

While random levels of impurities resulting in random poreformation at different sites is also possible the majority ofpores will be formed at the outermost surface resulting in a PSshell or sponge surrounding a solid Si core. In the case ofsample 2E-100 which has an average particle size of about 4–5mm then this porous outer shell can be up to 1 mm thick asobserved through FIB and SEM.19,48 In sample 2E-60 a similarpattern structure is observed but the porous outer shell isy600 nm thick for the same sized particles as sample2E-100.19,48 This is due to the time difference in the etchingprocesses. Secondary electrons (SE) are those electrons emittedby the specimen which are having energies in the range of 0 to50 eV. Due to these weak energies of SE, they are limited tovery small depth called escape depth below the surface of theparticle.49 Hence Secondary Electron Imaging (SEI) is highlydependent on surface topography of the material.

6. Conclusions

The porous silicon (PS) used in this study is comprised ofmetallurgical grade PS particles that contain a porous sponge-like boundary up to 1 mm thick with a solid silicon core as aresult of the etching process. CBED analysis on individualparticles has indicated no shift in HOLZ Lines when differentareas of the particles of the as-received unetched material isanalysed. Shifts in these HOLZ lines occur on etched samples,but diffraction contrast imaging indicates the presence of smallcrystallites (y1 nm) on the surface of etched particles which are

likely to have led to the HOLZ line shifts. The discovery of thesesmall crystallites etched from the pores of the Si particles has notpreviously been reported in these materials, but is backed up inthis paper through micro Raman measurements.

XPS, ICP and XRF data confirms the presence of impuritiesin the Si particles analysed, while the TOF-SIMS data gives amap of the elemental distributions which has not beenobserved previously. It is clear that the porosity formationmechanisms for MGSi when chemically etched using themethod outlined previously is due to the galvanic displace-ment reactions between the metal ions (impurities) at thesurface of the MGSi particles and the chemical etchantcomposition used. The Raman data highlights a free energyreduction in the samples post-chemical etching, which mayindicate strain relaxation within individual particles due to thereduction of impurity levels.

Acknowledgements

The Authors would like to acknowledge the financial supportof Enterprise Ireland, Vesta Sciences (EI IP 2007 0380 Vesta/UL) and PRTLI cycle 4. The support of Vesta Sciences and DrShanti Subramanian for providing PS samples, XRF and ICPdata. The authors would also like to thank Paula Olsthoorn,Gaye Hanrahan, Calum Dickinson, Wynette Redington, andShohei Nakahara for analytical results and useful discussion.

References

1 A. Uhlir, Electrolytic shaping of germanium and silicon,Bell Syst. Tech. J., 1956, 35, 333–347.

2 E. Xifre Perez, Design, fabrication and characterization ofporous silicon multilayer optical devices, PhD Thesis,Universitat Rovira I Virgili, 2007.

3 R. Guerrero-Lemus, F. A. Ben-Hander, J. L. G. Fierro,C. Hernandez-Rodrıguez and a. J. M. Martınez-Duart,Compositional and photoluminescent properties of anodi-cally and stain etched porous silicon, Phys. Stat. Sol., 2003,197(a), 137–143.

4 L. T. Canham, Silicon quantum wire array fabrication byelectrochemical and chemical dissolution of wafers, Appl.Phys. Lett., 1990, 57, 1046.

5 V. Parkhutik, Analysis of Publications on Porous Silicon:From Photoluminescence to Biology, J. Porous Mater., 2000,7(1), 363–366.

6 V. H. F. Lehman, Formation Mechanism and Properties ofElectrochemically Etched Trenches in n-Type Silicon, J.Electrochem. Soc., 1990, 137(2), 653–659.

7 V. U. G. Lehman, Porous silicon formation: A quantumwire effect, Appl. Phys. Lett., 1991, 58, 856.

8 C. O’Dwyer, D. N. Buckley, D. Sutton and S. B. Newcomb,Anodic Formation Characterization of Nanoporous InP inAqueous KOH Electrolytes, J. Electrochem. Soc., 2006,153(12), G1039–G1046.

9 C. O’Dwyer, D. N. Buckley, D. Sutton, M. Serantoni and S.B. Newcomb, An Investigation by AFM and TEM of the

This journal is � The Royal Society of Chemistry 2013 RSC Adv.

RSC Advances Paper

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online

Mechanism of Anodic Formation of Nanoporosity in n-InPin KOH, J. Electrochem. Soc., 2007, 154(2), H78–H85.

10 C. O’Dwyer, D. N. Buckley and S. B. Newcomb,Simultaneous Observation of Current Oscillations andPorous Film Growth during Anodization of InP, Langmuir,2005, 21, 8090–8095.

11 V. Lehman, Electrochemistry of Silicon, Wiley-VCH,Weinhem, 2002.

12 M. J. Sailor, Practice. Preparation, Characterization andApplications, in Porous Silicon, Wiley-VCH, Weinhem, 2011.

13 C. Becker, L. Currano and W. Churaman, Characterizationand Improvements to Porous Silicon Processing forNanoenergetics, Storming Media-Pentagon Reports, 2008, 20.

14 E. J. Anglin, Porous silicon in drug delivery devices andmaterials, Adv. Drug Delivery Rev., 2008, 60(11), 1266–1277.

15 S. Subramanian, S. Y. Limaye and D. Farrell, 2006 SiliconNanosponge Particles, Ireland Patent WO/2006/121870, 16November 2006.

16 K. Kareh, A. Ghahremaninezhad and E. Asselin,Electrochemical properties of metallurgical-grade siliconin hydrochloric acid, Electrochim. Acta, 2009, 54(26),6548–6553.

17 S. Subramanian, T. Tiegs and S. Limaye, Paper presented atthe proceedings of the Army Science Conference, Orlando, 1–4December 2008.

18 M. Watanabe, Microsc. Microanal., 2007, 13, 962–963.19 E. Chadwick, S. Beloshapkin and D. A. Tanner,

Microstructural Characterisation of Metallurgical GradePorous Silicon Nanosponge Particles, Journal of MaterialsScience, 2012, 45, 2396–2404.

20 M. De Graef, Introduction to Conventional TransmissionElectron Microscopy, Cambridge University Press, 2003.

21 K. Peng, Motility of Metal Nanoparticles in Silicon andInduced Anisotropic Silicon Etching, Adv. Funct. Mater.,2008, 18(19), 3026–3035.

22 L. M. Sorokin, Structural and electrophysical properties ofa nanocomposite based upon the Si–SiO2 system,Microscopy of semiconducting materials, 2005, 107, 327–332.

23 S. Chattopadhyay, X. Li and P. W. Bohn, In-plane control ofmorphology and tunable photoluminescence in poroussilicon produced by metal-assisted electroless chemicaletching, J. Appl. Phys., 2002, 91(9), 6134–6140.

24 R. Douani, Formation of aligned silicon-nanowire onsilicon in aqueous HF/(AgNO3 + Na2S2O8) solution, Appl.Surf. Sci., 2008, 254(22), 7219–7222.

25 S. Yae, Formation of porous silicon by metal particleenhanced chemical etching in HF solution and itsapplication for efficient solarcells, Electrochem. Commun.,2003, 5(8), 632–636.

26 N. Megouda, Au-assisted electroless etching of silicon inaqueous HF/H2O2 solution, Appl. Surf. Sci., 2009, 255(12),6210–6216.

27 I. Solomon, Intense photoluminescence of thin films ofporous hydrogenated microcrystalline silicon, J. Appl.Phys., 2008, 103(8), 083108–4.

28 D. Dimova-Malinovska, Preparation of thin porous siliconlayers by stain etching, Thin Solid Films, 1997, 297(1–2), 9–12.

29 Z. C. Feng and A. T. S. Wee, Multi-Technique study ofporous silicon membranes by Raman Scattering, FTIR,XPS, AES, and SIMS, in POROUS SILICON, R. T. Z. C. Feng,Editor, 1994, pp. 175–194.

30 J. F. W. F. S. Moulder, P. E. Sobol and K. D. Bomben,Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corporation Physical Electronics Division, 1992.

31 NIST-XPS Database, Version 4.1, (http://srdata.nist.gov/xps/).32 P. G. Abramof, Investigation of nanostructured porous

silicon by Raman spectroscopy and atomic force micro-scopy, J. Non-Cryst. Solids, 2004, 338–340, 139–142.

33 M. Ohmukai, The effects of chemical etching of porous siliconon Raman spectra, Czech. J. Phys., 2004, 54(7), 781–784.

34 P. P. L. Z. Sui and P. Irving, Raman analysis of light-emittingporous silicon, 1992.

35 R. S. Dubey and D. K. Gautam, Synthesis andCharacterization of Nanocrystalline Porous Silicon Layerfor Solar Cells Applications, Journal of Optoelectronic andBiomedical Materials, 2009, 1(1), 8–1.

36 F. P. S. Kozlowski and W. Lang, Micro Raman Investigationsand a Structure Model for Electroluminescent Porousn-Silicon, Porous Silicon, 1994, 149–171.

37 J. C. P. M. M. Tsang, F. Dacol and J. O. Chu, J. Appl. Phys.,1994, 75, 8098.

38 K. W. Kolasinski, Silicon nanostructures from electrolesselectrochemical etching, Curr. Opin. Solid State Mater. Sci.,2005, 9, 73–83.

39 M. I. J. Beale, J. D. Benjamin, M. J. Uren, N. G. Chew and A.G. Cullis, An experimental and theoretical study of theformation and microstructure of porous silicon, Journal ofCrystal Growth, 1985, 73(3), 622–636.

40 G. Oskam, J. G. Long, A. Natarajan and P. C. Searson,Electrochemical deposition of metals onto silicon, 1997, 31,1927–1949.

41 R. L. Smith and S. D. Collins, Porous silicon formationmechanisms, J. Appl. Phys., 1992, 71, R1.

42 A. Halimaoui, Porous Silicon Formation by Anodisation, inProperties of Porous Silicon, L. Canham, Editor, INSPEC-TheInstitution of Electrical Engineers, 1997, pp. 12–22.

43 J. L. Coffer, Porous Silicon Formation by Stain Etching, inProperties of Porous Silicon, L. Canham, Editor, INSPEC-TheInstitution of Electrical Engineers, 1997, pp. 23–29.

44 L. Canham, Properties of Porous Silicon, INSPEC-TheInstitution of Electrical Engineers, 1997.

45 O. Bisi, S. Ossicini and L. Pavesi, Porous silicon: a quantumsponge structure for silicon based optoelectronics, Surf.Sci. Rep., 2000, 38(1–3), 1–126.

46 U. Gosele and V. Lehmann, Porous silicon quantumsponge structures: Formation mechanism, preperationmethods and some properties, in Porous Silicon, Z. C.Feng and R. Tsu, Editors, World Scientific Publishing Co,New York, 1995, pp. 17–39.

47 F. Beck and H. Gerischer, Redoxvorgange an Germanium-Elektroden. Zeitschrift fur Elektrochemie, Berichte derBunsengesellschaft fur physikalische Chemie, 1959, 63(8),943–950.

48 E. G. Chadwick, D. A. Tanner, S. Beloshapkin, H. Zhang andG. Behan, High Resolution Microscopical Analysis ofMetallurgical Grade Nanoporous Silicon Particles, NUI Galway-UL Alliance First Annual Engineering and Informatics ResearchConference, National University of Ireland, 2011.

49 R. F. Egerton, Physical principles of Electron Microscopy-AnIntroduction to TEM, SEM, and AEM, Springer, USA, 1st edn,2005, p. 202.

RSC Adv. This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

02

Aug

ust 2

013.

Dow

nloa

ded

by U

nive

rsity

of

Lim

eric

k on

09/

09/2

013

10:5

9:38

. View Article Online